Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
Journal of Glaciology
Article
Cite this article: Montagnat M, Bourcier M,
Philip A, Bons PD, Bauer CC, Deconinck P,
Hereil P (2021). Texture characterization of
some large hailstones with an automated
technique. Journal of Glaciology 1–15. https://
doi.org/10.1017/jog.2021.66
Received: 13 January 2021
Revised: 17 May 2021
Accepted: 18 May 2021
Keywords:
Hail; ice crystal studies; ice in the atmosphere
Author for correspondence:
Maurine Montagnat,
E-mail: maurine.montagnat@univ-grenoble-
alpes.fr
© The Author(s), 2021. Published by
Cambridge University Press. This is an Open
Access article, distributed under the terms of
the Creative Commons Attribution-
NonCommercial-ShareAlike licence (http://
creativecommons.org/licenses/by-nc-sa/4.0/),
which permits non-commercial re-use,
distribution, and reproduction in any medium,
provided the same Creative Commons licence
is included and the original work is properly
cited. The written permission of Cambridge
University Press must be obtained for
commercial re-use.
cambridge.org/jog
Texture characterization of some large
hailstones with an automated technique
Maurine Montagnat1,2 , Mathieu Bourcier1, Armelle Philip1, Paul D. Bons3,4 ,
Catherine C. Bauer3, Paul Deconinck5and Pierre Hereil5
1
Univ. Grenoble Alpes, CNRS, IGE, F-38000 Grenoble, France;
2
Univ. Grenoble Alpes, Université de Toulouse, Météo
France, CNRS, CNRM, Centre d’Etudes de la Neige, F-38000 Grenoble, France;
3
Department of Geosciences,
Eberhard Karls University Tübingen, Tübingen, Germany;
4
China University of Geosciences, Beijing, China and
5
Thiot Ingénierie, Puybrun, France
Abstract
Hailstone structures have been studied for over a century, but so far mainly by manual optical
means. This paper presents new texture and microstructure data (i.e. crystal lattice orientations,
grain sizes and shapes) measured with an Automatic Ice Texture Analyzer, which gives access to
high spatial and angular resolutions. The hailstones show two main characteristics: (1) they are
structured with several concentric layers composed of alternating fine equiaxed grains and coarse
elongated and radially oriented grains, and (2) they show two texture types with c-axes oriented
either parallel or perpendicular to the radial direction. Such textures are compared with the ones
observed in lake S1 and S2 ices, respectively. The S1 texture type (with c-axes parallel to the col-
umnar crystals that grew in the radial direction) may result from epitaxial growth from a poly-
crystalline embryo, while the S2 texture (c-axes in the plane perpendicular to the column
direction) may result from the growth from an embryo made of a few crystals with mainly
one crystallographic orientation. Our novel high-resolution maps and measurements of both
microstructure and texture may help to shed new light on the long-term discussion on the growth
mechanisms of large hailstones.
Introduction
Hail fall during thunderstorms can create severe damage and losses to agriculture, car holders
and real estate. In terms of insured losses, the two hailstorms in Germany on 27–28 July 2013
were the costliest natural disasters worldwide in that year, with total damage estimated at more
than 6 billion Euros (Kunz and Kugel, 2015; Puskeiler and others, 2016). Aircraft are also
strongly impacted by hailstones when passing through hailstone regions and clouds, poten-
tially causing serious accidents.
Interest in hailstone microstructures is not new (Crammer, 1903) and was first motivated
by the understanding of hailstone formation in clouds, based, for example, on isotopic studies
associated with direct microstructural observations (Schuma, 1938; List, 1960b; Knight and
Knight, 1968; Jouzel and others, 1975; Macklin and others, 1976; Macklin, 1977). More
recently, a large range of studies focused on understanding and modeling of the impact of hail-
stones on structures (see e.g. Anghileri and others, 2005; Kim and Keune, 2007; Deconinck,
2019). In most of these studies, modeling approaches are validated with laboratory experi-
ments in which hailstones are replicated by artificial ice samples, sometimes spherical, with
microstructures that are mostly far from realistic (Combescure and others, 2011; Guégan
and others, 2011; Pernas-Sánchez and others, 2015).
Grain size and texture (here used to denote the distribution of crystallographic orientations)
are known to influence the mechanical behavior of ice, in particular in the static brittle regime
(see Schulson and Duval, 2009, for a review). Impact of hailstones occurs in a dynamical
regime (at strain rates higher than several s
−1
). For most materials, it is assumed that porosity
(density, shape and size distribution of pores) controls fracture propagation during dynamic
behavior (Forquin and Erzar, 2010) and this is also true for ice (Georges and others, 2021).
However, very few studies have considered the effect of microstructure and texture under
such conditions. This may be due to the lack of accurate data and the difficulty to design hail-
like microstructures and textures.
Hailstone microstructure results from the formation history in cumulonimbus clouds.
Hailstones formation is related to convective instabilities, atmospheric humidity, and to a
lesser extent wind and aerosol content in the cloud. From Knight and others (1975) and
Macklin and others (1976) early studies, formation conditions can be constrained indirectly
by measuring the isotopic signatures in the different ice layers forming a hailstone. Some
authors favor growth of hailstones during re-injection of embryos and updraft, for example
based on isotopic measurements (Jouzel and others, 1975) or modeling (Browning and
Foote, 1976). Large hailstones may also form along a simple trajectory with no recirculation
(Knight and others, 1975). A recent modeling study revisit the impact of updraft trajectory
conditions on the hailstone size (Kumjian and Lombardo, 2020).
Hailstones form out of embryos that mostly consist of graupel (List, 1958), but they may
also originate from frozen water drops (Rasmussen and Heymsfield, 1987). Depending on
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
temperature and aerosol content, hailstone growth can result from
the rapid freezing of supercooled water (Macklin, 1977), but also
from the freezing of a liquid shell and all intermediate cases.
Hailstone growth is separated into two main regimes, wet and
dry growth. During dry growth, accretion of water occurs by
immediate freezing upon contact. The minimum temperature of
growth is constrained by the lowest supercooled water tempera-
ture of −36°C. Hail growth is termed wet when a fraction of
water remains liquid at the surface of the growing hail or graupel.
Wet growth was extensively studied in the laboratory, with pion-
eer work to be found in List (1960b) and Macklin and Ludlam
(1961). Wind-tunnel experiments enabled Lesins and List
(1986) to show that wet growth can be further subdivided, ranging
from moist, to spongy and finally soaked, depending on the tem-
perature, water content and gyration rate, all leading to different
microstructures. Under extreme conditions, hail microstructure
can result in dendritic structures such as observed by Knight
and Knight (2005) on large hailstones. Aerosols may also come
into play during growth (Ilotoviz and others, 2016). High concen-
trations of aerosols increase the number of available supercooled
droplets and favor accretion of large hailstones. A lower concen-
tration of aerosols will, in contrast, favor the formation of more
numerous but smaller hailstones by water droplet freezing.
Gyration during falling impacts the final shape of hailstones,
and leads to approximately oblate spheroids, instead of spheres
that are often used for modeling (see e.g. Knight and Knight,
1970; García-García and List, 1992).
As a consequence of their complex formation processes, large
hailstones mainly resemble oblate spheroids, and often consist of
alternating layers of ‘white’and ‘dark’ice. The difference between
these layers is directly related to porosity and grain size. White
(opaque) ice layers, composed of very small grains surrounded
by tiny air bubbles, grow under dry conditions (Macklin, 1977).
Transparent layers (that appear ‘dark’in comparison with white
layers) form under wet conditions during which a liquid layer
remains in contact with the ice surface over longer periods. As
a consequence, grains are generally larger and fewer bubbles are
trapped. A hailstone may encounter both growth conditions
repeatedly, depending mainly on its falling or rising velocity, tem-
perature and liquid water content in the cloud. When analyzing
hailstones, it should be borne in mind that the microstructure
can change due to recrystallization after hailstone formation,
especially when reaching higher temperatures on the ground
(Macklin, 1977; Knight and Knight, 2005).
Although hailstone microstructures have been extensively
studied and modeled, only very few studies exist that investi-
gated the crystallographic orientations of ice crystals in large
hailstones. Knight and Knight (1968) used surface etching to
determine the crystallographic c-axis orientations of large col-
umnar grains in the outer shells of hailstones. With the low reso-
lution of their method, only the larger crystals could be
evaluated and represented by one single orientation measure-
ment each. The c-axes were found to be oriented radially, in
the growth direction, and from this observation Knight and
Knight (1968) disproved the hypothesis of a spongy growth
for the studied large hailstones.
In this study, we present the first high-resolution microstruc-
ture and texture measurements on large hailstones (a few centi-
meters in diameter) that were gathered from the town of
Tübingen in southwest of Germany, during a violent storm of
28 July 2013 (Kunz and Kugel, 2015; Puskeiler and others,
2016) and in the southwest of France on 31 August 2015. Our
data may offer complementary constraints for modeling-based
approaches to simulate and predict hailstone formation and
impact. They may also help to improve the design of ice-growing
techniques to better mimic natural hailstones.
Characterization of hailstone textures
Hailstone collection
Hailstones from Germany were collected by PDB and CCB about
a quarter of an hour after the storm that took place in the south-
west German town of Tübingen, on 28 July 2013. They were piled
up on the street and balcony, while the temperature had dropped
to ∼20°C. They were immediately stored at ∼−32°C until three of
them (samples G1–3) were analyzed.
The French hailstones were collected by volunteers from the
ANELFA association (Association Nationale d’Etude et de Lutte
contre les Fleaux Atmospheriques, http://www.anelfa.asso.fr/), in
the southwest of France on 31 August 2015. They were put in sili-
con oil, and kept at −28°C before being send to IGE (Institut of
Geosciences of Environment) where they were stored at −10°C
until four of them were analyzed (samples F1–4). One of these
large hailstones is illustrated in Figure 1.
Method
Ice Ih, the main component of hailstones, is characterized by a
hexagonal crystallographic structure and an optical birefringence.
The long axis of the hexagonal structure, the c-axis, is also the
optical axis. Observation of thin sections of ice under cross-
polarized light has therefore been used for a long time to deter-
mine the orientation of c-axes of individual crystals. Classically
performed by manual rotation and inclination of the thin section
on a Rigsby stage (Rigsby, 1951), analyses are today automatic,
fast and of high accuracy thanks to Automatic Ice Texture
Analyzers (AITA). The AITA used in this study is inherited
from developments by Russell-Head and collaborators (http://
www.russellheadinstruments.com), from the early 2000
(Russell-Head and Wilson, 2001; Wilson and others, 2003,
2007), based on the idea that rotation and inclination of the
thin section can be automatized, and that the c-axes distribution
can be derived from a set of digital images of the thin section
taken from these different angles and inclinations. A full descrip-
tion of the automated procedure of the AITA version used here
can be found in Peternell and others (2009), and we will provide
the main technical details here.
In the AITA version used in this study, the thin section is posi-
tioned on a stage that moves horizontally to move the 10 mm×10
mm camera field of view over the full section. Over this field of
view, stacks of images are taken from eight inclined quasi-
monochromatic light emitting diodes (LEDs) and three vertical
white LEDs that mimic the rotation of the section between
cross polarizers. These stacks of images are treated by AITA soft-
ware to determine the extinction planes that contains the c-axis.
From this the c-axis orientation is calculated for each pixel in
the image and reported as (1) its azimuth (measured clockwise
from north, between 0° and 360°), (2) its colatitude or plunge
(between 0° and 90°, corresponding to a c-axis parallel and per-
pendicular to the section plane, using a lower-hemisphere projec-
tion) and (3) a quality factor (QF) that estimates the accuracy of
the extinction planes matching (between 0 and 100). Our equip-
ment enables c-axis orientation measurements at a spatial reso-
lution down to 5 μm. In this study, the resolution was set to 10
μm as a necessary compromise between the size of the data files
and the minimum measurable grain size (which is also con-
strained by a minimum thin section thickness). The angular reso-
lution is ∼3° (Peternell and others, 2011). In agreement with the
detailed AITA technical study performed by Peternell and others
(2009,2011), the QF threshold was set at 70 or 75 in order to
secure the reliability of the c-axis orientation measurements. In
doing so, pixels from grain boundaries and bubbles are automat-
ically removed.
2 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
All results obtained with the AITA are analyzed with an
in-house python script (https://github.com/ThomasChauve/aita).
This script draws map of the thin section, using a c-axis orienta-
tion based color lookup table (shown as inset in the figures). It
also provides a pole figure that corresponds to a lower-hemisphere
equal-area stereographic projection of the c-axis orientations
(Fig. 2). Owing to the large number of pixel-data available, only
a random selection of 10 000 c-axis orientations (above the QF
threshold) is displayed. The color scale for pole figures represents
the (normalized) width of the density probability function
(obtained from a kernel density estimation). Pole figures are
shown in the (x,y) plane that corresponds to the thin section
plane. The z-axis is perpendicular to the thin section plane and
to the figures, unless stated differently.
A thin section is a 2-D sample of a 3-D object and therefore
may provide biased information. In order to test the representa-
tiveness of the thin-section based data, two perpendicular section
of one hailstone were analyzed. The orientation maps and pole
figures (Fig. 3) of these two thin sections show similar microstruc-
tures and textures. This confirms, at least qualitatively, the con-
centric nature of the texture and microstructure, and that the
2-D analysis of thin sections provides meaningful results.
Most of the studied hailstone microstructures are composed of
quasi-ellipsoidal layers. We used individual elliptical masks to
provide and visualize data of individual layers (Figs 5,6).
Opaque or cloudy layers reflect the porosity of hailstones (see
for instance Fig. 1). In the orientation maps, porosity appears as
white areas without any successful orientation measurements,
similarly to grain boundaries or mis-indexed areas (Figs 2,3).
In order to reveal and quantify the porosity, X-ray micro-
computed tomography (microCT) proved to be perfectly suited
for porous ice and snow (see Coléou and others, 2001, for pio-
neering work). Here, we present a preliminary study made on a
40×20×20 mm
3
sample from the French hailstone F2. The sample
was scanned with the Easytom XL nano tomograph from RX
Solutions of the CMTC (Consortium des Moyens
Technologiques Communs) platform at the Grenoble-INP
(Institut National Polytechnique) (Fig. 4), maintained at −10°C
in a cold cell during the acquisition (Burr and others, 2018).
One acquisition was done to cover the whole sample volume
with a (cubic) voxel size of 5 μm, and another one to focus on
a smaller volume and increase the resolution to a voxel size of
2μm. The high contrast between air and ice, which ensures an
easy threshold-based segmentation of the 3-D binary images,
facilitates the reconstruction of the 3-D microstructure from the
raw grayscale density images. Figure 4 shows a 3-D view of the
reconstructed porous microstructure for the two volumes (large
volume and low resolution in Fig. 4b, small volume and higher
resolution in Fig. 4c).
Since a full statistical evaluation of the porosity of hailstones is
beyond the scope of the paper, we here only provide preliminary
data as an illustration. The porosity is estimated to be ∼1.5 vol%
in the larger sample scanned (Fig. 4b) and ∼2.2 vol% in the
‘zoomed’part (Fig. 4c). In this last volume, two very large
pores account for 26.5% of the total porosity, each one having a
volume of ∼0.1 mm
3
. Pores smaller than the voxel size cannot
be observed which underestimates the porosity. Porosity appears
very heterogeneously distributed within the different layers with
small and large pores co-existing in the coarse-grained layers.
Observations and analyses
Seven large hailstones of diameter between 3.5 and 5 cm were
studied, four from the southwest of France storm (samples F1–4)
and three from the German storm (samples G1–3). No clear dis-
tinction in the microstructures and textures could be made related
to the geographic location of the presented hailstones. However, we
have observed two types of dominant textures for the hailstones
studied, which we illustrate in the following and for which we
will propose interpretations in the next section. The illustration
will be supported visually by detailed characterizations of two rep-
resentative hailstones. Observations done on the other hailstones
can be found in the Appendix.
In the following we will refer to ‘core’when mentioning the
inner part of the hailstone, since we cannot accurately assert
that this corresponds to the embryo. Hypotheses about a related
embryo type will nevertheless be given.
Hailstone F1 is oblate spheroidal with a major axis of ∼5.2 cm,
a minor axis of ∼4 cm and a thickness of ∼1.5 cm (Fig. 1). The
AITA analysis (Fig. 2) shows a concentric zonation with varia-
tions in grain size and shape. The pole figure of the entire hail-
stone shows a girdle-type texture, with most c-axes oriented
close to the plane of the thin section.
Figure 5 presents four individual concentric layers that were
isolated from the F1 hailstone together with their corresponding
pole figures. The core of the hailstone (Fig. 5a) is composed of
a large grain surrounded by several smaller grains. Since the
blue and red colors correspond to almost the same orientation
but with opposite azimuth (see color wheel in Fig. 5), the core
of hailstone F1 is composed of grains with almost identical
c-axis orientations. From this observation, we can hypothesize
that the embryo resulted from a frozen droplet that further
grew under wet conditions.
The first noticeable layer (Fig. 5b) is composed of large grains
elongated in the radial direction, which is the assumed growth dir-
ection (Knight and Knight, 2005). A close look at the intersection
between the core and this layer in Figure 2 reveals that this layer
starts from a fined-grained area, and a porous area with large elon-
gated pores is observed at the transition. In this coarse-grained layer,
the texture is anisotropic, with c-axes oriented preferentially in a gir-
dle perpendicular to the thin section plane. The analysis of several
individual grain orientations indicates that most of them have
their c-axis oriented perpendicular to their growth direction (or
their long axis), as denoted by the black arrows in Figure 5.
The second layer (Fig. 5c) is a fine-grained layer that corre-
sponds to the opaque, porous layer of Figure 1. This layer also
has girdle-type c-axis distribution, with c-axes preferentially
aligned close to the plane of the thin section. The slight misalign-
ment could be due to the fact that the thin section was not cut
exactly through the center of the hailstone. The large number of
small grains, many of them too small to be oriented, prevent us
Fig. 1. Flat hailstone from the south west of France (F1) observed under polarized
light.
Journal of Glaciology 3
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
from deciphering the orientation of the c-axes relative to the
growth direction. The last layer (Fig. 5d) is composed of large,
more equiaxed grains, also characterized by a girdle-type c-axis
distribution. In this layer, some of the large grains have their
c-axes oriented in the growth direction, and some are oriented
perpendicular to this direction. No clear trend can be extracted
owing to the limited number of grains in the layer.
The second hailstone used as an illustration is the German
hailstone G1 (Fig. 2). One core and two individual layers could
be clearly distinguished (Fig. 6). The last layer, being too complex,
will not be commented here. The core (Fig. 6a) is composed of
many small equiaxed grains (about a hundred micrometers in
diameter). Such a microstructure would be in favor of a high por-
osity graupel embryo. Both the first and second layers (Figs 6b,c)
are composed of radially aligned elongated grains, but grains are
smaller in the second layer. The three parts (core and layers) show
a distribution of c-axes characterized by a girdle oriented in the
plane of the thin section. Colors in the microstructure images
of the two external layers show a remarkable match with the
color distribution of the color wheel, which highlights the radial
orientations of the c-axes.
Discussion
In the following, we will discuss our observations in light of the
knowledge of the relationship between growth mechanisms and
resulting textures and microstructures in lake ice. Such a choice
will later be discussed regarding the hypotheses of wet growth
mechanisms. There are two types of lake ice, termed S1 and S2,
each with their own growth mechanism.
In terms of texture, S1 ice has its c-axes preferentially oriented
in the direction of growth, while S2 ice is characterized by c-axes
oriented in the plane perpendicular to the growth direction. In
terms of microstructure, both types mostly show grains elongated
in the direction of growth, known as columnar ice. Michel and
Ramseier (1971) classified these two types of ice based on the
departing point of their growth, called the ‘primary’layer, that
acts as seed. The classification provides four types of primary
layers. One with seed-controlled c-axes orientations, resulting in
S1-type ice, and the other three forming S2-type ice, owing to
growth-controlled c-axis orientations. In order to induce the
growth of S1 columnar ice, the primary or seed layer has to
form slowly on still water, in such a way that grains, all with ver-
tical c-axes, are covering the surface. On the contrary, the primary
Fig. 2. Illustration of AITA data obtained from German hailstone G1 (a) and French hailstone F1 (b). For each hailstone: Left: orientation color-coded image
obtained after a filtering with QF set to 75. White areas are excluded for further analyses. The orientation color-code is given by the color wheel on the bottom
left of the images (lower-hemisphere, equal area stereographic projection). Scale is mm. Right: c-axis orientations plotted on a pole figure. The color-code corre-
sponds to the density of pixels. The (x,y) plane is the plane of the thin section.
4 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
(seed) layers that lead to S2 ice formation are characterized by
randomly oriented small grains, that result either from falling
snow, or from an agitated surface. S1-type ice can therefore
only grow if grains with c-axes oriented in the growth direction
are present at the onset of growth, and under conditions quiet
enough to prevent subsequent perturbation of the growth front.
All hailstones studied here have in common that they are com-
posed of concentric layers with different grain size, morphology
and texture. Our high-resolution measurements of c-axis
orientations and microstructures reveal two specific textures. The
first type is represented by the F1 hailstone, Figure 5 (see also
Figs 9 and 10 in the Appendix for a similar hailstone F4). It
includes samples composed of a nearly single-crystalline core, sur-
rounded by a layer with large elongated grains with c-axes oriented
tangentially, perpendicular to the growth direction. These specific
crystalline orientations and morphology are comparable to the
S2-type lake ice described before. The c-axis orientations would
result, similarly to S2 ice, from a selective growth process, known
Fig. 3. Microstructures and pole figures of two
perpendicular thin sections of the French hail-
stone F2.
Fig. 4. Micro-computed X-ray tomography visualization
of the 3-D porosity in one part of the French hailstone
F2 (a) with a resolution of 5 μm (b) and within a smaller
volume, with a resolution of 2 μm (c). Pores are shown in
color, ice is transparent.
Journal of Glaciology 5
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
as competitive growth or survival of the fastest (Bons and Bons,
2003) also observed in icicles extracted from frozen ice falls
(Montagnat and others, 2010). Layers similar to the second layer
of hailstone F1 (Fig. 5c) are commonly observed is the hailstones
studied here. It is composed of fine grains with c-axes oriented
with a relative isotropy in the thin section plane. A closer look at
a few individual grains shows that both radial and tangential
orientations are present (not shown here). Based on the grain
size observation, we may assume that this layer is the result of
the rapid freezing on the hailstone surface of supercooled droplets
present in the cloud, very likely under dry conditions. Our texture
observations seem to show that grain orientations present in this
fine-grained layer could have impacted the crystalline orientations
of the following layer by playing the role of seeds. Indeed the last
Fig. 5. Extracted microstructures of the layers
observed in the French hailstone F1 with the corre-
sponding pole figures. Core layer (top) to external
layer (bottom). Black arrows sketch the orientation of
a few selected c-axes to help with the analyses. The
scale of the orientation color-coded map is in milli-
meter. The color code is provided by the color wheel.
6 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
layer shows large grains with some c-axes oriented along the
growth direction, as observed in the G1 hailstone.
The second type of hailstone, G1, Figure 6 (see also Figs 7,8in
the Appendix for a similar hailstone F3) also presents several layers
but with another arrangement. The large elongated grains are char-
acterized by c-axes oriented in the direction of growth. Contrary to
the F1-type hailstone, the microstructure of the core is composed of
small grains. From this core, the growth process in the subsequent
layer seems to have favored a specific texture with c-axes oriented
radially along the growth direction. Such a specific texture, that is
not favored by growth, can be compared to the one observed in
S1-type lake ice described before and that was shown to result
from an oriented seed-based growth, under still conditions.
Finally, the French F3 hailstone (see Figs 7,8in the Appendix)
is an interesting illustration of the complexity of the layer forma-
tion. The core is composed of a limited number of grains, with
one main orientation. But this core is surrounded by a large
layer of very fine grains. This layer seems to play the role of
seed for the subsequent layers. This latter showing an S1-type tex-
ture, with radial c-axes. In turn, once interrupted in their growth
process, the resulting rather large grains give way to a layer with
an S2-type texture.
Crystal growth during hailstone formation could therefore fol-
low mechanisms similar to those well documented for S1 and S2
lake ices. The conditions necessary to meet S1-type ice (c-axes
oriented in the growth direction already present in the seed,
and still growth conditions) seem to be satisfied in the G1-type
hailstone within which the girdle-type texture observed in the
fine-grained core (very likely resulting from a low porosity graupel
embryo) provides the necessary seed orientations for the growth
of grains with c-axes in the radial direction. The large dimension
of these radially oriented elongated grains, and the low porosity
(transparent ice), must result from relatively undisturbed and
slow growth conditions (Knight and Knight, 2005).
On the contrary, these conditions do not seem to apply to hail-
stones similar to F1 since the core is formed of a limited number
Fig. 6. Extracted microstructures of the layers observed
in the German hailstone G1 with the corresponding
pole figures. Core layer (top) to external layer (bottom).
Black arrows sketch the orientations of a few selected
c-axes to help with the analyses. The scale of the orien-
tation color-coded map is in millimeter. The color code
is provided by the color wheel.
Journal of Glaciology 7
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
of crystals with similar orientations (very likely resulting, in this
case, from a freezing droplet embryo). Among all hailstones stud-
ied here (see also the Appendix), those with an S2-type texture
had cores containing only a few large grains.
Spongy wet growth has been identified as a relevant mechan-
ism for the formation of large hailstones for a long time (see for
instance the pioneer work by List, 1960a,1960b, Knight and
Knight, 1968). More recently, List (2014) and Phillips and others
(2014) have suggested convincing modeling approaches that help
to reinforce the likelihood of such a mechanism operating. During
wet spongy growth, the hailstone is surrounded (partly or com-
pletely) by a liquid water layer resulting from its interaction
with droplets in the cloud. The analysis performed in our study
is therefore compatible with a wet and maybe spongy growth con-
figuration, as the presence of water in contact with the growing ice
surface is required. This analysis is not relevant for dry growth
conditions that results in very fine-grained ice layers.
In their pioneer work, Knight and Knight (1968) aimed at cor-
relating the likelihood of spongy growth with the characteristics of
c-axis preferred orientations. The authors measured c-axis orienta-
tions by means of etching on the largest crystals from various nat-
ural large hailstones. Based on the hypothesis that hailstones could
result from a spongy growth or not, the authors suggested that
spongy hailstones were characterized by large elongated grains
with a tangential c-axis preferred orientations, while large elongated
grains in non-spongy hailstones would show more radial c-axis
orientations. Knight and Knight (1968) concluded that, based on
samples from five different storms, natural hailstones were likely
not resulting from a spongy growth. Thanks to the high resolution
of the crystallographic orientation measurements performed here we
areabletoprovideamoreaccuratepicture.Bothradialc-axes and
c-axes perpendicular to the growth direction are observed in indi-
vidual hailstones (e.g. type F1, Fig. 5), and texture is therefore not
an unequivocal indicator for spongy growth, unless various growth
processes can occur during the formation of a single large hailstone.
Our observations therefore highlight that, in order to complement
recent studies on large hailstone formation (List, 2014; Phillips
and others, 2014; Ilotoviz and others, 2016), crystallographic orien-
tation characterizations by means of high-resolution c-axes mea-
surements are an interesting and potential useful additional tool.
Mechanical behavior of hailstones is of interest for predicting,
and preventing, damage caused by hail impact. Hail impact occurs
at relatively high strain rate, therefore in the dynamic regime (strain
rate above ∼10 s
−1
). Under these conditions, as mentioned in the
Introduction, no study exists that takes into account microstructure
and texture as factors that influence the mechanical response. Under
quasi-static conditions, many studies exist, and show a limited effect
of grain size on fracture toughness, with variations ranging from
∼70 kPa m
1/2
for 9 mm grain size to ∼90 kPa m
1/2
for 2 mm grain
size, and no clear effect of texture (see Schulson and Duval, 2009,
for a review). On the contrary, porosity does impact the fracture
toughness under quasi-static conditions, with a decrease of ∼25%
when the porosity increases from 0 to ∼15 vol% in fresh granular
ice (Smith and others, 1990). In the dynamic loading range (strain
rates between 20 and 130 s
−1
), Georges and others (2021)showeda
strong effect of porosity on the velocity-dependent tensile strength,
which decreases by ∼25% when the porosity increases from 1 to
∼10 vol%. Few of the ice-impact studies performed so far considered
the microstructure and texture of the laboratory grown samples,
which may explain the highly variable reported strength measure-
ments (see e.g. Shazly and others, 2009; Combescure and others,
2011; Tippmann and others, 2013). We therefore suggest to pay
more attention to grain sizes and porosity in order to design statis-
tical microstructures as close as possible to the hailstone microstruc-
tures as characterized here. In addition, the role to texture on the
mechanical strength under dynamic conditions relevant for hail
impact remains to be studied. It is only under these conditions
that the modeling of hail impact on structures will be efficient
(see e.g. Meyers, 1994; Erzar and Forquin, 2010).
Conclusions
This paper presented high-resolution characterizations of the micro-
structureandtexture(c-axis orientations) of several large hailstones
originating from two storms, one that took place in France in 2015,
and the other one in Germany in 2013. High spatial and angular
resolution observations of textures were made possible by thin sec-
tion analyses with an Automatic Ice Texture Analyser. We observed
strong microstructure similarities between all hailstones studied, with
a quasi-spherical layered structure, in which layers of small equiaxed
grains alternate with layers of large elongated (columnar) grains.
Despite these strong similarities, we recognized two main types of
hailstones, each with a specific crystallographic texture within suc-
cessive layers. One type is mostly made of layers with c-axes oriented
radially, in the direction of the large grain columns. The other type
presents some layers with c-axisorientationsperpendiculartothe
radial (and growth) direction. These different types seem to result
from the specific properties of the core or the inner contiguous
layer (in terms of orientation and number of grains), and are com-
pared to the extensively studied S1 and S2 types of lake ice, providing
insight into their formation history.
So far very few studies exist on the effect of texture and micro-
structures on the brittle strength, to predict their effect on the
dynamic response of hailstone ice and resulting damage would
be speculative. This study reveals the full microstructure complex-
ity, including layered structure, crystal shapes, orientations and
porosity within hailstones that should be taken into account to
correctly extrapolate laboratory experiments to natural conditions.
Data availability. Data are available on Zenodo database under the reference
(Montagnat and others, 2020)(https://doi.org/10.5281/zenodo.3938956). Data
treatment can be performed by using the freely accessible Python toolbox
developed by Thomas Chauve (https://github.com/ThomasChauve/aita).
Acknowledgments. The authors acknowledge association ANELFA and
volunteers who collected the French hailstones in the southwest of
France. A. Burr is greatly acknowledged for his help during tomography mea-
surements, and T. Chauve for his support on the Python tools he developed.
Author contributions. MM wrote the article and organized the interactions
between co-authors. MM made some of the texture measurements and ana-
lyses. MB wrote the first draft of the article and made the main measurements
and analyses (including microCT). AP participated in the writing and in
the measurements and analyses of the microCT data. PB participated in the
writing and analyses of the data. PB and CB provided the German hailstones.
PD and PH provided the French hailstones and provided some financial sup-
port. We acknowledge support by the Open Access Publishing Fund of the
University of Tübingen.
Financial support. Support was provided by CNRS INSIS and INSU institutes.
The French Direction Générale de l’Armement (DGA) supported this research
through the framework of project RAPID 142906128. Labex OSUG@2020 partly
supported this study in the frame of the project ANR10Labex56. We acknowledge
support by the Open Access Publishing Fund of the University of Tübingen.
Conflict of interest. The authors declare that the research was conducted in
the absence of any commercial or financial relationships that could be con-
strued as a potential conflict of interest.
References
Anghileri M, Castelletti LM, Invernizzi F and Mascheroni M (2005) A sur-
vey of numerical models for hail impact analysis using explicit finite elem-
ent codes. International Journal of Impact Engineering 31(8), 929–944.
8 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Bons AJ and Bons PD (2003) The development of oblique preferred orienta-
tions in zeolite films and membranes. Microporous and Mesoporous
Materials 62(1), 9–16. doi: 10.1016/S1387-1811(03)00384-6.
Browning KA and Foote GB (1976) Airflow and hail growth in supercell storms
and some implications for hail suppression. Quarterly Journal of the Royal
Meteorological Society 102(433), 499–533. doi: 10.1002/qj.49710243303.
Burr A and 5 others (2018) Pore morphology of polar firn around closure
revealed by X-ray tomography. The Cryosphere 12(7), 2481–2500. doi: 10.
5194/tc-12-2481-2018.
Coléou C, Lesaffre B, Brzoska JB, Ludwig W and Boller E (2001)
Three-dimensional snow images by X-ray microtomography. Annals of
Glaciology 32,75–81. doi: 10.3189/172756401781819418.
Combescure A, Chuzel-Marmot Y and Fabis J (2011) Experimental study of
high-velocity impact and fracture of ice. International Journal of Solids and
Structures 48(20), 2779–2790. doi: 10.1016/j.ijsolstr.2011.05.028.
Crammer H (1903) Eis-und gletscherstudien. Neuer Jahrbuch für mineralogie,
geologie und paleontologie 18,57–116.
Deconinck P (2019) Development of a hailstone substitute for representative
impact tests. Technical report. doi: 10.4271/2019-01-1942.
Erzar B and Forquin P (2010) An experimental method to determine the ten-
sile strength of concrete at high rates of strain. Experimental Mechanics 50
(7), 941–955. doi: 10.1007/s11340-009-9284-z.
Forquin P and Erzar B (2010) Dynamic fragmentation process in concrete
under impact and spalling tests. International Journal of Fracture 163(1),
193–215. doi: 10.1007/s10704-009-9419-3.
García-García F and List R (1992) Laboratory measurements and para-
meterizations of supercooled water skin temperatures and bulk properties
of gyrating hailstones. Journal of Atmospheric Sciences 49(22), 2058–2073.
doi: 10.1175/1520-0469(1992)049<2058:LMAPOS>2.0.CO;2.
Georges D, Saletti D, Montagnat M, Forquin P and Hagenmuller P (2021)
Influence of porosity on ice dynamic tensile behavior as assessed by spalling
tests. Journal of Dynamic Behavior of Materials. doi: 10.1007/s40870-021-
00300-z.
Guégan P and 6 others (2011) Experimental investigation of the kinematics of
post-impact ice fragments. International Journal of Impact Engineering 38
(10), 786–795.
Ilotoviz E, Khain AP, Benmoshe N, Phillips VTJ and Ryzhkov AV (2016)
Effect of aerosols on freezing drops, hail, and precipitation in a midlatitude
storm. Journal of the Atmospheric Sciences 73(1), 109–144. doi: 10.1175/
JAS-D-14-0155.1.
Jouzel J, Merlivat L and Roth E (1975) Isotopic study of hail. Journal of
Geophysical Research 80(36), 5015–5030. doi: 10.1029/JC080i036p05015.
Kim H and Keune JN (2007) Compressive strength of ice at impact strain
rates. Journal of Materials Science 42(8), 2802–2806. doi: 10.1007/
s10853-006-1376-x.
Knight CA, Ehhalt DH, Roper N and Knight NC (1975) Radial and tangential
variation of deuterium in hailstones. Journal of Atmospheric Sciences 32(10),
1990–2000. doi: 10.1175/1520-0469(1975)032<1990:RATVOD>2.0.CO;2.
Knight CA and Knight NC (1968) Spongy hailstone growth criteria
I. Orientation fabrics. Journal of the Atmospheric Sciences 25(3), 445–452.
Knight CA and Knight NC (1970) Hailstone embryos. Journal of Atmospheric
Sciences 27(4), 659–666. doi: 10.1175/1520-0469(1970)027<0659:HE>2.0.CO;2.
Knight CA and Knight NC (2005) Very large hailstones from Aurora,
Nebraska. Bulletin of the American Meteorological Society 86(12), 1773–
1781. doi: 10.1175/BAMS-86-12-1773.
Kumjian MR and Lombardo K (2020) A hail growth trajectory model for
exploring the environmental controls on hail size: model physics and idea-
lized tests. Journal of the Atmospheric Sciences 77(8), 2765–2791. doi: 10.
1175/JAS-D-20-0016.1.
Kunz M and Kugel PI (2015) Detection of hail signatures from single-
polarization C-band radar reflectivity. Atmospheric Research 153, 565–577.
Lesins GB and List R (1986) Sponginess and drop shedding of gyrating hail-
stones in a pressure-controlled icing wind tunnel. Journal of Atmospheric
Sciences 43(23), 2813–2825. doi: 10.1175/1520-0469(1986)043<2813:
SADSOG>2.0.CO;2.
List R (1958) Kennzeichen atmosphärischer eispartikeln. Zeitschrift für ange-
wandte Mathematik und Physik ZAMP 9(2), 180–192. doi: 10.1007/
BF01600631.
List R (1960a) Design and operation of the Swiss hail tunnel. Washington DC
American Geophysical Union Geophysical Monograph Series 5, 310–316.
doi: 10.1029/GM005p0310.
List R (1960b) New developments in hail research. Science 132(3434), 1091–
1098), ISSN 00368075, 10959203.
List R (2014) New hailstone physics. Part I: heat and mass transfer (HMT) and
growth. Journal of the Atmospheric Sciences 71(4), 1508–1520.
Macklin W (1977) The characteristics of natural hailstones and their interpret-
ation. Hail: A Review of Hail Science and Hail Suppression, Meteorological
Monographs 38,65–88.
Macklin WC, Carras JN and Rye PJ (1976) The interpretation of the crystal-
line and air bubble structures of hailstones. Quarterly Journal of the Royal
Meteorological Society 102(431), 25–44. doi: 10.1002/qj.49710243103.
Macklin WC and Ludlam FH (1961) The fallspeeds of hailstones. Quarterly
Journal of the Royal Meteorological Society 87(371), 72–81. doi: 10.1002/
qj.49708737109.
Meyers MA (1994) Dynamic Behavior of Materials. USA: Wiley-Interscience.
Michel B and Ramseier R (1971) Classification of river and lake ice. Canadian
Geotechnical Journal 8(1).
Montagnat M and 6 others (2010) Waterfall ice: formation, structure and evo-
lution. Journal of Glaciology 56(196), 225–234.
Montagnat M and 5 others (2020) On the birth of structural and crystallo-
graphic fabric signals in polar snow: a case study from the EastGRIP snow-
pack. Frontiers in Earth Science 8, 365, ISSN 2296-6463. doi: 10.3389/feart.
2020.00365.
Pernas-Sánchez J, Artero-Guerrero J, Varas D and López-Puente J (2015)
Analysis of ice impact process at high velocity. Experimental Mechanics
55(9), 1669–1679.
Peternell M, Kohlmann F, Wilson CJ, Seiler C and Gleadow AJ (2009) A
new approach to crystallographic orientation measurement for apatite fis-
sion track analysis: effects of crystal morphology and implications for auto-
mation. Chemical Geology 265(3–4), 527–539, ISSN 0009-2541. doi: 10.
1016/j.chemgeo.2009.05.021.
Peternell M, Russell-Head DS and Wilson CJL (2011) A technique for
recording polycrystalline structure and orientation during in situ deformation
cycles of rock analogues using an automated fabric analyser. Journal of
Microscopy 242(2), 181–188. doi: 10.1111/j.1365-2818.2010.03456.x.
Phillips VTJ, Khain A, Benmoshe N and Ilotoviz E (2014) Theory of time-
dependent freezing. Part I: description of scheme for wet growth of hail.
Journal of the Atmospheric Sciences 71(12), 4527–4557. doi: 10.1175/
JAS-D-13-0375.1.
Puskeiler M, Kunz M and Schmidberger M (2016) Hail statistics for
Germany derived from single-polarization radar data. Atmospheric
Research 178, 459–470.
Rasmussen RM and Heymsfield AJ (1987) Melting and shedding of
graupel and hail. Part I: model physics. Journal of Atmospheric
Sciences 44(19), 2754–2763. doi: 10.1175/1520-0469(1987)044<2754:
MASOGA>2.0.CO;2.
Rigsby GP (1951) Crystal fabric studies on Emmons Glacier Mount Rainier,
Washington. The Journal of Geology 59(6), 590–598. doi: 10.1086/625914.
Russell-Head DS and Wilson CJL (2001) Automated fabric analyser system
for quartz and ice. In Geological Society of Australia Abstracts, vol. 64,
159–159, Geological Society of Australia; 1999.
Schulson EM and Duval P (2009) Creep and Fracture of Ice. Cambridge, UK:
Cambridge University Press.
Schuma TEW (1938) The theory of hailstone formation. Quarterly Journal of the
Royal Meteorological Society 64(273), 3–21. doi: 10.1002/qj.49706427303.
Shazly M, Prakash V and Lerch BA (2009) High strain-rate behavior of ice
under uniaxial compression. International Journal of Solids and Structures
46(6), 1499–1515. doi: 10.1016/j.ijsolstr.2008.11.020.
Smith T, Schulson M and Schulson E (1990) The fracture toughness of por-
ous ice with and without particles. In Proceedings, Ninth International
Conference on Offshore Mechanics and Arctic Engineering, 241–246.
Tippmann JD, Kim H and Rhymer JD (2013) Experimentally validated strain
rate dependent material model for spherical ice impact simulation.
International Journal of Impact Engineering 57,43–54. doi: 10.1016/j.ijim-
peng.2013.01.013.
Wilson CJL, Russell-Head DS, Kunze K and Viola G (2007) The analysis of
quartz c-axis fabrics using a modified optical microscope. Journal of
Microscopy 227(1), 30–41. doi: 10.1111/j.1365-2818.2007.01784.x.
Wilson C, Russell-Head D and Sim H (2003) The application of an auto-
mated fabric analyzer system to the textural evolution of folded ice layers
in shear zones. Annals of Glaciology 37(1), 7–17. doi:10.3189/
172756403781815401.
Journal of Glaciology 9
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Appendix A
Figures 7 and 8show the microstructure of a French hailstone called F3. This
‘star geometry’is a typical geometry for wet growth (Knight and Knight, 2005).
The sample size was estimated by making the assumption of an oblate spher-
oidal shape. The large axis is ∼6.4 cm, the medium axis is ∼5.5 cm and the lit-
tle axis is ∼4.5 cm.
The latest French hailstone studied, called F4, is relatively spherical with a
diameter of ∼3.5 cm (Fig. 9). The sample is opaque which testifies of the
presence of a rather homogeneous porosity distributed in the hailstone. Its
microstructure and the corresponding pole figure are presented in Figures
9and 10.
The German hailstone called G2 is relatively complex, and the different
layers are uneasy to distinguish. However, we can observe four different layers
including first an ex-centered embryo with relatively large grains, a second
layer with relative large grains (∼1 mm in diameter), a third layer with small
equiaxed grains and a fourth layer with ellipsoidal grains elongated in the dir-
ection of growth. The global texture of this hailstone is of girdle-type, with
most c-axis orientations in the radial direction (Figs 11,12).
The last studied hailstone called G3 is a specific one where only two layers
could be identified (Fig. 13). One large layer with homogeneous equiaxed
grains and a second small layer with elongated grains. The first layer presents
pole figure with a ‘ring-like’texture, and the second one presents a preferential
orientation with two main poles.
Fig. 7. Orientation color-coded microstructure of the
French hailstone F3 and the corresponding pole figure.
For each hailstone: Left: orientation color-coded image
obtained after a filtering with QF set to 75. White areas
are excluded for further analyses. The orientation color-
code is given by the color wheel on the bottom left of
the images (lower-hemisphere, equal area stereographic
projection). Scale is mm. Right: c-axis orientations plot-
ted on a pole figure. The color-code corresponds to
the density of pixels. The (x,y) plane is the plane of
the thin section.
10 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Fig. 8. Orientation color-coded representation of the
different layers of the French hailstone F3 and corre-
sponding pole figures. See Figure 7 for color-code
and scale explanation.
Journal of Glaciology 11
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Fig. 9. Orientation color-coded microstructure of the
French hailstone F4 and the corresponding pole figure.
See Figure 7 for color-code and scale explanation.
Fig. 11. Orientation color-coded microstructure
of the German hailstone G2 and pole figure.
See Figure 7 for color-code and scale
explanation.
12 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Fig. 10. Orientation color-coded representation of the
different layers of the French hailstone F4 and corre-
sponding pole figures. See Figure 7 for color-code and
scale explanation.
Journal of Glaciology 13
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Fig. 12. Orientation color-coded representation of the dif-
ferent layers of the German hailstone G2 and corresponding
pole figures. See Figure 7 for color-code and scale
explanation.
14 Maurine Montagnat and others
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.
Fig. 13. Orientation color-coded representation
of the layers of the German hailstone G3 and
corresponding pole figures. See Figure 7 for
color-code and scale explanation.
Journal of Glaciology 15
Downloaded from https://www.cambridge.org/core. 16 Jun 2021 at 17:05:12, subject to the Cambridge Core terms of use.